in Navigating the Storm
Ken Birman, Robbert van Renesse, Werner Vogels Dept. of Computer Science Cornell University
Autonomic Computing
A challenge both technically but also from a business perspective
Suppose someone came along with a major advance in autonomic technology Would we really adopt it?
Most COTS platforms have their issues, yet work well enough Killer app for autonomic market?
AMS 2003 Autonomic Computing 2
June 24, 2003
Lacking autonomic tools
Web Services availability problems
WS goal is to give instant response from legacy back-end systems that may run in batches Web site not responding What to do?
Grid applications may find it hard to track down appropriately configured nodes
Issue is one of scale arises with lots of nodes, not just a handful And the grid needs to administer itself
3
June 24, 2003
Autonomic tools could solve such AMS 2003 Autonomic Computing
Why might such tools help?
In these (and other new areas, such as sensor networks and large-scale data mining) applications are expected to
Seek out appropriate resources Self-repair after disruption Do so automatically Match work load to available capacity
Were solving new problems with new tools, not trying to fix old, very complex problems
AMS 2003 Autonomic Computing 4
June 24, 2003
Grid Computing example
Doctor Kildare works in the Cayuga Medical Center in Ithaca
She does a CT scan on a patient with a head injury Wants to run a new generation of computationally intensive tools to assist her in the procedure to stabilize this patient Then may need to find a specialist who can consult on this situation
June 24, 2003
AMS 2003 Autonomic Computing
5
Grid Computing example
In the past, Cayuga Medical needed a dedicated computer for this task With Grid Computing
Should be able to grab resources on a farm (perhaps operated by IBM or some other company), run the application, get the results over the network Cayuga Medical spends far less June 24, 2003 6 money AMS 2003 Autonomic Computing
To run reliably must
Automatically find machines on which the right software is running, with the right revision level for this use
Perhaps there are intermediate data sets that we need to access Perhaps that software, in turn, will need to use other services (like a specific database system) Communication latencies need to be low
How could grid ever know enough about her application to solve such problems?
AMS 2003 Autonomic Computing 7
June 24, 2003
Grid capability implied?
The grid should have a way to
Allow an application the grid has never seen before to view a pool of hardware in a new and application-specific way that lets the application express a preference to run on appropriate nodes
AMS 2003 Autonomic Computing 8
Sounds like a database problem
June 24, 2003
Beyond basics
We also need to sense and adapt if disruption occurs
For example, a failure or a network problem this are more common than one might think
And need to parameterize the run based on the configuration we find
AMS 2003 Autonomic Computing 9
June 24, 2003
We lack the right tools!
Today, our applications navigate in the dark
They lack a way to find things They lack a way to sense system state There are no rules for adaptation, if/when needed
In effect: We are starting to build very big systems, yet doing so in the usual client-server manner This denies applications any information about system state, configuration, June 24, 2003 10 loads, etc AMS 2003 Autonomic Computing
Astrolabe
Astrolabe is our information monitoring and replication architecture It has three components
Mariner: a novel kind of virtual database Multicast: for faster few to many data transfer patterns Kelips: A fast lookup mechanism June 24, 2003 AMS 2003 Autonomic Computing
11
First focus on Mariner
Mariners role is to track information residing at a vast number of sources Structured to look like a database Approach: peer to peer gossip. Basically, each machine has a piece of a jigsaw puzzle. Assemble it on the fly.
AMS 2003 Autonomic Computing 13
June 24, 2003
Mariner in a single domain
Name swift falcon cardinal
Load 2.0 1.8 2.1 1.9 3.1 1.5 1.1 0.8 0.9 4.5 2.7 3.6 5.3
Weblogi c? 0 1 1
SMTP ? 1 0 0
Word Version 6.2 4.1 6.0
Row can have many columns Total size should be k-bytes, not megabytes Configuration certificate determines what data AMS 2003 Autonomic Computing table (and14 is pulled into the June 24, 2003
So how does it work?
Each computer has
Its own row Replicas of some objects (configuration certificate, other rows, etc)
Periodically, but at a fixed rate, pick a friend pseudo-randomly and exchange states efficiently (bound the size of data exchanged)
States converge exponentially rapidly. Loads are low and constant and protocol is robust against all sorts of disruptions!
AMS 2003 Autonomic Computing 15
June 24, 2003
State Merge: Core of Mariner epidemic
Name
Time
Load
Weblogic ? 0 1 1
SMTP?
Word Versi on 6.2 4.1 6.0
swift falcon cardinal
2011 1971 2004
2.0 1.5 4.5
1 0 0
swift.cs.cornell.edu
Name swift falcon cardinal
Time 2003 1976 2201
Load .67 2.7 3.5
Weblogi c? 0 1 1
SMTP? 1 0 1
Word Version 6.2 4.1 6.0
cardinal.cs.cornell.edu
June 24, 2003
AMS 2003 Autonomic Computing
16
State Merge: Core of Mariner epidemic
Name
Time
Load
Weblogic ? 0 1 1
SMTP?
Word Versi on 6.2 4.1 6.0
swift falcon cardinal
2011 1971 2004
2.0 1.5 4.5
1 0 0
swift.cs.cornell.edu
swift
2011
2.0
cardinal Name swift falcon cardinal Time 2003 1976 2201 Load .67 2.7 3.5 Weblogi c? 0 1 1 SMTP? 1 0 1 Word Version 6.2 4.1 6.0
2201
3.5
cardinal.cs.cornell.edu
June 24, 2003
AMS 2003 Autonomic Computing
17
State Merge: Core of Mariner epidemic
Name
Time
Load
Weblogic ? 0 1 1
SMTP?
Word Versi on 6.2 4.1 6.0
swift falcon cardinal
2011 1971 2201
2.0 1.5 3.5
1 0 0
swift.cs.cornell.edu
Name swift falcon cardinal
Time 2011 1976 2201
Load 2.0 2.7 3.5
Weblogi c? 0 1 1
SMTP? 1 0 1
Word Version 6.2 4.1 6.0
cardinal.cs.cornell.edu
June 24, 2003
AMS 2003 Autonomic Computing
18
Observations
Merge protocol has constant cost
One message sent, received (on avg) per unit time. The data changes slowly, so no need to run it quickly we usually run it every five seconds or so Information spreads in O(log N) time
But this assumes bounded region size
In Mariner, we limit them to 50-100 24, June 2003 AMS 2003 Autonomic Computing 19 rows
Scaling up and up
With a stack of domains, we dont want every system to see every domain
Cost would be huge
Name Time Load Weblogi SMTP? Word c? Name Time Load Weblogi SMTP? Version Word Version Name 2011 Time 2.0 Load 0 c? Weblogi 1 SMTP? 6.2 Word swift c? Version swift Name 2011 Time 2.0 Load 0 Weblogi 1 SMTP? 6.2 Word falcon 1976 2.7 1 4.1 c? 0 Version swift Name 2011 Time 2.0 Load 0 Weblogi 1 SMTP? 6.2 Word falcon 1976 2.7 1 4.1 c? 0 Version SMTP? 6.2 Word cardinal swift Name 2011 Time 2.0 1Load 0 Weblogi 1 6.0 2201 3.5 1 falcon 1976 2.7 1 4.1 c? 0 Version SMTP? 6.2 Word cardinal swift Name 2011 Time 2.0 1Load 0 Weblogi 1 6.0 2201 3.5 1 falcon 1976 2.7 1 4.1 c? 0 Version cardinal swift 2201 3.5 2011 2.0 1 01 1 6.0 6.2 falcon 1976 2.7 1 0 4.1 cardinal swift 2201 3.5 2011 2.0 1 01 1 6.0 6.2 falcon 1976 2.7 1 0 4.1 cardinal 2201 3.5 1 1 6.0 falcon 1976 2.7 1 0 4.1 cardinal 2201 3.5 1 1 6.0 cardinal 2201 3.5 1 1 6.0
So instead, well see a summary
cardinal.cs.cornell.edu
June 24, 2003
AMS 2003 Autonomic Computing
20
Build a hierarchy using a P2P protocol that assembles the puzzle without any servers
Dynamically changing query output is visible system-wide
Name SF NJ Paris Avg Load 2.6 1.8 3.1
WL contact 123.45.61. 3 127.16.77. 6 14.66.71.8
SMTP contact 123.45.61.17 127.16.77.11 14.66.71.12
SQL query summarizes data
Weblogic? 0 0 1 SMTP? 0 1 0 Word Version 4.5 6.2 6.2
Name swift falcon cardinal
Load 2.0 1.5 4.5
Weblogic? 0 1 1
SMTP? 1 0 0
Word Version 6.2 4.1 6.0
Name gazelle zebra gnu
Load 1.7 3.2 .5
San Francisco
June 24, 2003
New Jersey
AMS 2003 Autonomic Computing 21
(1) Query goes out (2) Compute locally (3) results flow to top level of the hierarchy
Name
Avg Load 2.6 1.8 3.1
WL contact 123.45.61. 3 127.16.77. 6 14.66.71.8
SMTP contact 123.45.61.17 127.16.77.11 14.66.71.12
1 3
Name swift falcon cardinal Load 2.0 1.5 4.5 Weblogic? 0 1 1 SMTP? 1 0 0
SF NJ Paris
1 3
Weblogic? 0 0 1 SMTP? 0 1 0 Word Version 4.5 6.2 6.2
Word Version 6.2 4.1 6.0
Name gazelle zebra gnu
Load 1.7 3.2 .5
2
2
San Francisco
June 24, 2003
New Jersey
AMS 2003 Autonomic Computing 22
Hierarchy is virtual data is replicated
Name SF NJ Paris
Avg Load 2.6 1.8 3.1
WL contact 123.45.61. 3 127.16.77. 6 14.66.71.8
SMTP contact 123.45.61.17 127.16.77.11 14.66.71.12
Name swift falcon cardinal
Load 2.0 1.5 4.5
Weblogic? 0 1 1
SMTP? 1 0 0
Word Version 6.2 4.1 6.0
Name gazelle zebra gnu
Load 1.7 3.2 .5
Weblogic? 0 0 1
SMTP? 0 1 0
Word Version 4.5 6.2 6.2
San Francisco
June 24, 2003
New Jersey
AMS 2003 Autonomic Computing 23
Hierarchy is virtual data is replicated
Name SF NJ Paris
Avg Load 2.6 1.8 3.1
WL contact 123.45.61. 3 127.16.77. 6 14.66.71.8
SMTP contact 123.45.61.17 127.16.77.11 14.66.71.12
Name swift falcon cardinal
Load 2.0 1.5 4.5
Weblogic? 0 1 1
SMTP? 1 0 0
Word Version 6.2 4.1 6.0
Name gazelle zebra gnu
Load 1.7 3.2 .5
Weblogic? 0 0 1
SMTP? 0 1 0
Word Version 4.5 6.2 6.2
San Francisco
June 24, 2003
New Jersey
AMS 2003 Autonomic Computing 24
Mariner solves our problem!
A flexible, user-programmable mechanism
Where can I find a display device suitable for displaying these NMR test results? Are any specialists in epilepsy available to consult on this ER admission? Find 12 idle computers with copies of the NMR-3D package to which I can download a 20MB dataset rapidly Is there a free office where I could use a computer and a phone?
Think of aggregation functions as small agents that look for information
AMS 2003 Autonomic Computing 25
June 24, 2003
Data Mining
Quite a hot area, usually done by collecting information to a centralized node, then querying within that node Mariner is doing the comparable thing, but computing (query evaluation) occurs in a decentralized manner
June 24, 2003
...